Ship_Tilt_Detection/loss.py

133 lines
4.3 KiB
Python

import torch
import torch.nn as nn
import torch.nn.functional as F
class BCELoss(nn.Module):
def __init__(self):
super(BCELoss, self).__init__()
def _gather_feat(self, feat, ind, mask=None):
dim = feat.size(2)
ind = ind.unsqueeze(2).expand(ind.size(0), ind.size(1), dim)
feat = feat.gather(1, ind)
if mask is not None:
mask = mask.unsqueeze(2).expand_as(feat)
feat = feat[mask]
feat = feat.view(-1, dim)
return feat
def _tranpose_and_gather_feat(self, feat, ind):
feat = feat.permute(0, 2, 3, 1).contiguous()
feat = feat.view(feat.size(0), -1, feat.size(3))
feat = self._gather_feat(feat, ind)
return feat
def forward(self, output, mask, ind, target):
# torch.Size([1, 1, 152, 152])
# torch.Size([1, 500])
# torch.Size([1, 500])
# torch.Size([1, 500, 1])
pred = self._tranpose_and_gather_feat(output, ind) # torch.Size([1, 500, 1])
if mask.sum():
mask = mask.unsqueeze(2).expand_as(pred).bool()
loss = F.binary_cross_entropy(pred.masked_select(mask),
target.masked_select(mask),
reduction='mean')
return loss
else:
return 0.
class OffSmoothL1Loss(nn.Module):
def __init__(self):
super(OffSmoothL1Loss, self).__init__()
def _gather_feat(self, feat, ind, mask=None):
dim = feat.size(2)
ind = ind.unsqueeze(2).expand(ind.size(0), ind.size(1), dim)
feat = feat.gather(1, ind)
if mask is not None:
mask = mask.unsqueeze(2).expand_as(feat)
feat = feat[mask]
feat = feat.view(-1, dim)
return feat
def _tranpose_and_gather_feat(self, feat, ind):
feat = feat.permute(0, 2, 3, 1).contiguous()
feat = feat.view(feat.size(0), -1, feat.size(3))
feat = self._gather_feat(feat, ind)
return feat
def forward(self, output, mask, ind, target):
# torch.Size([1, 2, 152, 152])
# torch.Size([1, 500])
# torch.Size([1, 500])
# torch.Size([1, 500, 2])
pred = self._tranpose_and_gather_feat(output, ind) # torch.Size([1, 500, 2])
if mask.sum():
mask = mask.unsqueeze(2).expand_as(pred).bool()
loss = F.smooth_l1_loss(pred.masked_select(mask),
target.masked_select(mask),
reduction='mean')
return loss
else:
return 0.
class FocalLoss(nn.Module):
def __init__(self):
super(FocalLoss, self).__init__()
def forward(self, pred, gt):
pos_inds = gt.eq(1).float()
neg_inds = gt.lt(1).float()
neg_weights = torch.pow(1 - gt, 4)
loss = 0
pos_loss = torch.log(pred) * torch.pow(1 - pred, 2) * pos_inds
neg_loss = torch.log(1 - pred) * torch.pow(pred, 2) * neg_weights * neg_inds
num_pos = pos_inds.float().sum()
pos_loss = pos_loss.sum()
neg_loss = neg_loss.sum()
if num_pos == 0:
loss = loss - neg_loss
else:
loss = loss - (pos_loss + neg_loss) / num_pos
return loss
def isnan(x):
return x != x
class LossAll(torch.nn.Module):
def __init__(self):
super(LossAll, self).__init__()
self.L_hm = FocalLoss()
self.L_wh = OffSmoothL1Loss()
self.L_off = OffSmoothL1Loss()
self.L_cls_theta = BCELoss()
def forward(self, pr_decs, gt_batch):
hm_loss = self.L_hm(pr_decs['hm'], gt_batch['hm'])
wh_loss = self.L_wh(pr_decs['wh'], gt_batch['reg_mask'], gt_batch['ind'], gt_batch['wh'])
off_loss = self.L_off(pr_decs['reg'], gt_batch['reg_mask'], gt_batch['ind'], gt_batch['reg'])
## add
cls_theta_loss = self.L_cls_theta(pr_decs['cls_theta'], gt_batch['reg_mask'], gt_batch['ind'], gt_batch['cls_theta'])
if isnan(hm_loss) or isnan(wh_loss) or isnan(off_loss):
print('hm loss is {}'.format(hm_loss))
print('wh loss is {}'.format(wh_loss))
print('off loss is {}'.format(off_loss))
# print(hm_loss)
# print(wh_loss)
# print(off_loss)
# print(cls_theta_loss)
# print('-----------------')
loss = hm_loss + wh_loss + off_loss + cls_theta_loss
return loss